On Wed, 16 Aug 2023 19:12:25 -0400
Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:
> > > +
> > > + for (i = 0; i < size; i++) {
> > > + old[i] = copy[i];
> > > +
> > > + if (copy_traceeval_data(types + i, stats + i,
> > > + vals + i, copy + i))
> > > + goto fail;
> > > + }
> >
> > I think we still need to rip through old[] and free strings, and also call
>
> Yes, I forgot to add that :-p
>
> > .release on types that define it, probably via data_release().
> >
I realized I actually had this kind of backwards. On failure, we want to put
back old and remove the data added to vals (aka copy).
Here's what I'm doing:
for (i = 0; i < size; i++) {
old[i] = copy[i];
if (copy_traceeval_data(types + i, stats + i,
vals + i, copy + i))
goto fail;
}
data_release(size, old, types);
return 0;
fail:
/* Free the new values that were added */
data_release(i, copy, types);
/* Put back the old values */
for (i--; i >= 0; i--) {
copy_traceeval_data(types + i, NULL,
copy + i, old + i);
}
return -1;
-- Steve
|